Skip to main content
Incident intelligence/SS-IR-048CASE FILE OPEN
Symbolic editorial illustration for SS-IR-048SERVANTSTACK // INCIDENT INTELLIGENCEFORENSIC IMAGE // VERIFIED FRAME
SS-IR-048 // INCIDENT REPORTReported

Tesla

Full Self-Driving Drove Into Low-Visibility Crashes, Triggering a 2.4 Million-Vehicle Federal Probe

EXECUTIVE BRIEF

On October 17, 2024, the National Highway Traffic Safety Administration opened a preliminary evaluation into Tesla's "Full Self-Driving" (FSD) system covering approximately 2.4 million vehicles across model years 2016 through 2024.

AUTONOMOUS & AVIATION
FAILURE CHAINTRACE COMPLETE
  1. 01TRIGGEROn October 17, 2024, the National Highway Traffic Safety Administration opened a preliminary evaluation into Tesla's…
  2. 02MACHINE ACTIONAutonomous actor
  3. 03MISSING GATEExecution gate and human override
  4. 04IMPACTPhysical safety
01 // INCIDENT SUMMARY

The short version

On October 17, 2024, the National Highway Traffic Safety Administration opened a preliminary evaluation into Tesla's "Full Self-Driving" (FSD) system covering approximately 2.4 million vehicles across model years 2016 through 2024.

02 // KEY FACTS

Case telemetry

INCIDENT
SS-IR-048
DATE
October 17, 2024
SYSTEM
Tesla
LOCATION / SCOPE
United States (probe opened by NHTSA; fatal crash in Rimrock, Arizona)
EVIDENCE
Reported
AI ROLE
Autonomous actor
HARM
Physical safety
SOURCES
3 cited records
03ENTRY POINT // WHAT HAPPENED

The event

On October 17, 2024, the National Highway Traffic Safety Administration opened a preliminary evaluation into Tesla's "Full Self-Driving" (FSD) system covering approximately 2.4 million vehicles across model years 2016 through 2024. The probe followed four reported crashes in which FSD-equipped Teslas encountered reduced-visibility conditions -- sun glare, fog, and airborne dust. One of those crashes killed a pedestrian in Rimrock, Arizona, roughly 100 miles north of Phoenix, in November 2023, when a 2021 Tesla Model Y struck and killed a person on foot. A second crash caused an injury. NHTSA said it would examine FSD's ability to detect and respond appropriately to reduced roadway visibility, whether comparable crashes had occurred, and whether software updates had changed the system's behavior in those conditions.

04CAUSAL TRACE // AI'S ACTUAL ROLE

What the machine did

FSD is a driving-automation system that perceives the road through cameras and executes steering, braking, and acceleration on its own between driver interventions. Regulators flagged that the system's degradation-detection logic failed to recognize when its cameras were effectively blinded by glare, fog, or dust, and did not reliably warn the driver or hand control back before the vehicle drove into a hazard. The machine made real-time perception and motion decisions in conditions where its own sensing was compromised, with no enforced check that a competent human had confirmed the vehicle could actually see well enough to proceed -- it kept driving at machine confidence while operating on degraded input.

Autonomous actorAutomation was a causal participant—not a decorative label for the system around it.
05BLAST RADIUS // CONSEQUENCES

Where the failure landed

One pedestrian was killed and at least one person was injured across the four crashes that prompted the investigation. NHTSA opened a formal preliminary evaluation into roughly 2.4 million vehicles, exposing Tesla to a potential recall of its flagship driver-assistance feature. The probe was subsequently escalated in March 2026 to an engineering analysis and expanded to approximately 3.2 million vehicles -- the final regulatory step before a mandated recall -- and intensified scrutiny of Tesla's broader autonomy and robotaxi claims.

06 // EVIDENCE STATUS

Reported

Documented in the cited public record. Follow the sources for the precise evidentiary posture.

SOURCE RECORD UPDATED 2026-07-09

07 // SOURCE LEDGER

3 cited records

  1. 01
  2. 02
  3. 03
08CONTROL FAILURE // MISSING GOVERNANCE

Execution gate and human override

The failure pattern in this case: Autonomous high-consequence action.

09INTERVENTION POINT // HUMAN IN THE MIDDLE

The moment the path could change

A trained operator receives the evidence, owns the go/no-go decision, and retains an immediate override.

AI PROPOSES→HUMAN OWNS THE DECISION→SYSTEM EXECUTES
10CONTROL DEPLOYMENT // AUTHORITYGATE

High-consequence gate · human override

The AuthorityGate Operational Resilience framework requires a human SME validation gate on the operating-domain envelope before an autonomous control loop is permitted to act. Every release of perception-and-control software would have to pass a change-validation gate in which a qualified safety engineer signs off that the system can detect and correctly handle each declared environmental condition -- including degraded sensing from glare, fog, and dust -- and that a verified fail-safe (warn-and-handback or controlled slowdown) fires when sensing falls below a validated confidence threshold. Under that gate, FSD could not ship or stay enabled in low-visibility conditions it had not been validated to handle: an SME would have to attest that camera-blinding scenarios were covered, or the system would be constrained out of those conditions until they were. The gate forces a human to own the boundary between "the car can see" and "the car is guessing," which is exactly the boundary that failed here.

RELEVANT KEYSTONE CONTROLHuman-in-the-Loop ValidationHow high-risk actions route to a named subject-matter expert who owns the go or no-go decision.
12 // THE ALTERNATIVE

Autonomy is a design choice.

See the operating model that keeps AI useful while preserving human authority at consequential moments.

Compare AgenticAI and AugmentedAI →